Archive for June, 2013

Some interesting stuff for MondoRescue 3.0.4


Well at least I hope so !!

It took me much more time than I expected, and made me sleep at insane hours (yes really !), but I think that the beta that I built today under should be of interest for MondoRescue users.

Some new distros are now added to the list of the one I’m building for (Ubuntu 13.04, Debian 7.1, Mageia 3, Fedora 18, …) Usual job for me, but hey, still of interest for people wanting the latest one.

Speaking of latest Debian, MondoRescue should now work better on it thanks to the fixes around #699 (NFS and SSHFS support), softlink in LocateDeps which have now their target scanned, and some deblint issues removed, including right test binaries placement, and man pages for perl scripts. I have someone working now with me on the re-integration of MondoRescue into Debian, so this version should help him reducing the diffs with his tree, and allow for the next steps (on the packaging itself) to go forward for our goal.

Of course, some annoying bugs reported have been fixed. At least we all hope so 😉

  • grub2 support in #686 by adding grub.conf, grub.cfg and modified to (Thanks to Victor Gattegno),
  • Fix support for OBDR tapes connected to SCSI cards and addition of an obdr modsboot parameter if some additional drivers should be removed,
  • an particularly nasty bug (there since rev 1 !) on SLES 11 SP2 by removing a dd in mindi which was filling the target dir,
  • the full management of scratch and tmp dir has been rewritten, hopefully for a better implementation now and a full resolution of #640,
  • the removal of the notion of number of media (limited to 50 abusively). Cf: #706
  • the addition of a -F option which allows mondoarchive to fail at first interaction. Cf: #702
  • sparse file support by adding a CLI parameter -M to modify the biggie file size limit, by default 64 MB, and by using st_blocks instead of st_size. Cf: #700
  • Fix #679 again and #685 by reducing the list of possible kernels (Thanks to Victor Gattegno and Didier Diaz)
  • Fix #696 for VGs not having a ‘-‘ in their name such as multipath aliases
  • Fix #703 by removing the errctl field used in star and log all star msgs into mondorestore.log. star is now considered production level as afio

And some improvements in the doc such as:

  • Adds more info in mindi log file (df -i, df -a, grub/lilo conf files, fdisk -l)
  • Add an example to use a fixed IP configuration at boot time through mindi.conf
  • Be more precise in messages displayed at restore time when a non-rewinding device should be used
  • Document the new -F and -M options
  • Precise the usage conditions of the exclude option (differences between device vs directory exclusion)
  • Indicate the other new tools than mkinitrd to help rebuilding the boot env (Thanks to Philippe Lefevre)
  • Precising compression rates on GUI button to ease choice (0 is no compression)

Some ongoing efforts at this point would require your help, dear reader to give appropriate feedback to see if we are making progresses:

  • Fix #628 by avoiding to erase choices done for variable bkpinfo->isodir and g_isodir_device in iso_fiddly_bits later in the code
  • Fix #208 by adding cifs support with first modifications to add cifs support as an external FS

So please help us by downloading this beta, which is IMO near from being the next stable and report wheher it’s working fine for you, or whether it’s working really fine for you 😉

I have to say that most of these bugs wouldn’t have been fixed without the work I’ve done for HP’s customers recently, who are using extensively the tool, and were needing these fixes to have it fully working in their environment (RHEL and SLES in multiple versions on HP ProLiant servers). Pending their feedback on this version, I’ll declare it stable if they give me the green light, which should arrive soon now.

Bdale Garbee will soon need your help


TES 2009 - Bdale Garbee
For all of you, Debian fans, or Freedombox fans, former HP colleague, or even just if you enjoyed one of his numerous talk everywhere around the world, here is an opportunity to bring back to Bdale what he brought to Open Source since so many years. Bdale and his family lost their house in the fire which has devastated the Black Forest of Colorado.

Of course, nothing will be able to replace souvenir and memories lost due to the fire. But I hope our community can unite and help one of its most talented members feel the friendship we all have for him and his great work and attitude in the past.

I hope he will open a paypal account or any other similar way for us to give even a minimal amount to help him and his family pass the sad moments, and recover at least from a material point of view rapidly from the disaster. I hope he will receive from all over the world where he left such nice impressions after his conferences and talks (as in Canberra this year, in Brussels for the Fosdem or in Prague), the kind words he deserves and proposals of support from everybody in a situation to do it.

Write to him at to show that lots of people think to him and associate themselves with his pain. I’m sure it will be one of the manifestation of support he will appreciate. This great Open Source Ambassador will soon need your help. Show again the power of a community.

First OpenStack Meetup for Rhône Alpes


The first meetup for the new OpenStack regional group will be organized by HP at the HP/Intel Solution Center the 27 of June 2013.

We should have presentations about XLcloud from Bull, Crowbar from SUSE, oVirt & OpenStack from Red Hat and the testing of OpenStack by HP (all that should be confirmed soon)

Refer to (in french) for more details and registration for those of you around.

Join us at HP: 5, avenue Raymond Chanas, Eybens, France
Access B10 HP

Solutions Linux 2013 is over: smaller exhibition, crowded conferences


Since the first Linux Expo back in 1999, the event has changed, first growing a lot, and it is now going back to a smaller size, however, keeping what makes it very attractive, a combination of enterprise boths, community boths and free conferences which are a big success every year.

I passed most of my time of Tuesday having very interesting discussions and news from the french community, discovering the new e-Novance big OpenStack player (thanks Erwan), talking FLOSS history with Denis Bodor on the Linux Mag both, talking with friends at Combodo (TeamIP and their great iTOP FLOSS CMDB) or at normation (Rudder and CFengine), and getting the latest news from FusionDirectory, FusionInventory, AFUL or APRIL members, and other multiple communities.

I didn’t see time passed ! But I had to jump in the RER to attend, for once I could, the reheasal of the Mozart Requiem in Etampes directed by Jean Belliard ! You should attend their concert the 15th of June there. That was multiple great pleasures: First to sing this master piece under the direction of Jean, something I have not been able to do during the 12 years I sung at the choir (I also miss the mass from Bach !). And also because I could join my daughter and sing also with her in the choir, and that’s something I wish to everybody 😉 Of course, lots of discussions at then end of the rehearsal again, shoretened alas to be able to go back to Paris and attend the second day of Solutions Linux.

I first participated to a round table on Open Source Governance with Véronique Torner (she has a great knowledge of the FLOSS governance world and her comments were remarkable), Gaël Blondelle who also has a very interesting knowledge of industrial use cases of FLOSS, and of the Eclipse Foundation, Benjamin Jean, who is the legal person I prefer to work with, as he can really make you love legal and lawyers !, even when you’re a developer, and he has numerous examples very interesting to share, so I recommend listening to him. The round table was lead by Patrice Bertrand in front of some 40 people in the room, which is great for a hard topic like Open Source Governance.

Of course, it’s always too short, but I hope we showed how useful it is to control Open Source usage in enterprise, as well as SMBs, and that the goal is to expand usage, and get the most out of it, instead of creating FUD.

And after that, a sandwish and it ws time to lead the Sysadmin track with my colleague François Jeanmougin.

We had initially planned to have 4 hours, and thus gave 30 minutes to each of our 6 speakers, plus 1 hour for a round table. However, later on, we were restricted to 3 hours only. That’s an issue as this track used to have a full day some years ago. I decided that I didn’t want to cancel any speaker as their talks were indeed very valuable in our context of Data Center automation with/using FLOSS. And each speaker accepted to reduce their talk to 20′, which is really short, but forced them somewhere to go straight to their point and we thus had a dense but highly interesting track ! And extremely well followed as you can see below:

We started with Jonathan Clarke from Normation who did a global overview of what needs to be done to automate complex IT environement with FLOSS solutions. As he didn’t had eough time to develop the project he’s contributing to (Rudder) I really encourage you to take time to read more on it.
Jonathan Clarke (Normation)

Next speaker was logically Erwan Talloc from Combodo to present how we can today by combining multiple FLOSS tools, provide to enterprises an ITIL compliant FLOSS solution stack. Of course their iTop product is at the center, relevant of course for a CMDB. This stack is currently developped for answering a customer need, and hopefully, we’ll be able next year to present a concrete use case of this during the track.
Erwan Talloc (Combodo)

And when we speak of central management of systems, Directory is near to your mind, as well as to Benoit Mortier’s one (from Opensides), who thinks natively LDAP ! He presented FusionDirectory, complementary to all what was shown before, allowing to manage through OpenLDAP, using a very nice Web based interface servers, PCs, IP phones, or fax, as well as users, sysem deployment and much more.
Benoit Mortier (Opensides)

Going deeper in system management, I was really waiting the next speaker Frédéric Crozat from SUSE and his systemd subject. Too bad he had some issues with his laptop (btrfs related, not systemd related even if who knows ;-)) and could only make a short presentation. Systemd is supposed to colsve all the badness of sysVinit, speed, deps, … As I really hate systemd at the moment, for the multiple problems it created for me in my system managements, with in short, problems with chroots, problems with no X started when another unrelated issue happen, problem with memory consumption (44 MB for an init replacement and they say it’s not bloated !), problem with debugging requiring too many commands to pass to understand what is wrong, … I would have liked to ask many questions. For sure next time. Frédéric made his best to defend systemd but why are distros adopting it too early is something I have issue understandig. Sometimes I like the Debian approach !!
Frédéric Crozat (SuSE)

Always low in system management, we then had a talk on redundancy with load balancers and for load balancers, from Sébastien Rohaut (GFI) who covered of course ipvsadm and LVS, but also the various tools added around to make it always available, including his own makealive, giving users the possibility to avoid paying tons of euros for dedicated hardware platforms when this software soltion fits probably 80% of the needs.
Sébastien Rohaut (GFI)

Finally, Jean Gabès came on stage to explain to us what was new in Shinken. After many years of Nagios related confs, we had Jean for the first time 2 years ago. AsI found his solution interesting, I wanted to have him again presenting the evolution of monitoring and the constraints brought by a cloud context to that area.
Jean Gabès (Shinken)

Finally we made a 30′ question and answer session with the attendees, some speakers ( Jonathan Clarke, Erwan Talloc) and additional ones (Christophe Sautier (Objectif Libre), Guillaume Leccese (Oxalyde), Laurent Pinsivy (Méréthis)) and we had a great deal of questions, especially around the positioning of Data Center automation with regards to the cloud approach and some user experience feedback. Hopefully the audience was well informed and this session overall will contribute to increase FLOSS usage in data center even more.
Table Ronde: Jonathan Clarke (Normation), Christophe Sautier (Ob

So next year, hopefully, we’ll have more time (4 hours at least), more space (a 120 people room) but still the same energy and competence from speakers to have an always interesting track.